App Engagement: Metrics and How to Improve Them

App engagement is how often people use an app after installing it and how much they do when they are there. It is decided earlier than most teams think. In RevenueCat's 2026 data on subscription apps, about half of all paid conversions happen on the day of install, and so do 55% of the cancellations of three-day trials. Users make up their minds in the first session.

So this guide spends most of its time there. Push notifications, streaks and win-back campaigns all matter, but they work on the people who survived day zero. Fix the first ten minutes and everything after it gets easier.

Measure the right thing first

Do not count app opens. An open caused by a notification and abandoned two seconds later looks the same in that number as a finished workout.

Pick the one action your app exists for: a lesson completed, a meal logged, a message sent, an order placed. That is your core action. "Engaged" means doing it regularly, and every change you make should be judged by whether more users do it more often.

Then track these four, by the week users installed:

Session length is the metric to be careful with. Long sessions are good for a video app and bad for a taxi app. Engagement and retention are also not the same thing: someone who checks a bank balance monthly is retained and barely engaged. You want both numbers, because retention tells you there is a problem and activation usually tells you where.

Fix the first session

Write down every screen between opening the app for the first time and completing the core action. Count them. Then cut.

• Move sign-up after the first success. Let people do the thing, then ask them to create an account to save it. Asking first loses the people who were only curious, before they have seen a reason to stay.
• Replace the tutorial with a guided first action. Nobody remembers three swipeable slides. Everybody remembers doing the thing once with a hint.
• Ask personal questions only if the answers visibly change what comes next. A fitness app that asks your goal and then shows a plan built for it earns its questions. One that asks and shows everyone the same home screen has wasted five taps.
• Start with something filled in. An empty state is the worst screen in any app. Give a sample project, a recommended plan or a first lesson ready to play.
• Hold back every permission prompt until the moment the user needs it. Camera when they tap "scan". Notifications after the first success.

Measure the result as a funnel. If 100 people install, 40 finish onboarding and 15 complete the core action, your biggest leak is inside onboarding, and no notification strategy will fix it.

Notifications: earn the permission, then be useful

You cannot message anyone until they allow it. iOS has always asked, and Android has required a runtime permission since Android 13. On iOS the system dialog appears once. If the user taps "Don't Allow", getting them back means sending them to Settings. Android gives you slightly more room, but not much.

So do not show the system dialog cold. Show your own screen first, after the user has done something, saying exactly what they will get: "Want a reminder at 8am tomorrow?" with a yes and a not now. Only trigger the system prompt when they tap yes. People who would have refused tap "not now", and you keep the chance to ask again later.

Once you have permission:

• Send things the user would want to know. "Your order shipped", "Anna replied" and a reminder they set themselves are welcome. "We miss you!" is why people turn notifications off.
• Time reminders to when that user normally opens the app, not to when your marketing team is at work.
• Deep link to the exact screen. A notification about a reply that opens the home screen is a broken promise.
• Give people switches for each type. Someone who wants delivery updates and no offers will otherwise turn off everything.
• Watch the opt-out rate after every send. If it jumps, that message cost you a channel.

Stay visible without interrupting

Notifications are not the only way back into the app, and the alternatives do not need anyone's permission.

• A home screen widget keeps your app's most useful number in view all day: steps so far, budget left, next lesson. Each glance is a reminder you did not have to send.
• Live Activities on iOS show something in progress, such as a delivery, a timer or a match score, on the lock screen while it happens.
• A weekly email summary works for apps where progress builds up. It reaches people who have notifications off.
• In-app messages suit feature announcements, since they reach people who are already there. One at a time, and never on top of the core action.

Give people a reason to come back

Reminders do not create a reason to return. The product has to.

• Something of theirs lives in the app: history, saved items, a plan, a streak.
• Something changes: new content, a daily puzzle, updated prices, a friend's activity.
• Something is unfinished: a course at 40%, a goal two days from done.

Streaks work when they mark something the user cares about, and they backfire when missing one day wipes out months. Build in a way to recover, such as a freeze or a grace day, or the first broken streak becomes the day the user quits.

Know why people leave

Ask people who cancel a subscription why, and two answers dominate in every category: it costs too much (25 to 45% of cancellations) and they were not using it enough (26 to 40%). Technical problems are only 3 to 7%.

"Not using it enough" is an engagement failure you can see coming. A subscriber who has not opened the app for two weeks is at risk, so that is when to act: a message showing what they have built up, a suggestion that fits what they did before, or a lighter version of the habit. Waiting until the renewal date is too late.

Do still fix crashes. They are rarely the stated reason for leaving, but on Android they cost you in another way: Google says apps over its bad behavior thresholds for crashes and freezes are "likely to be less discoverable" on Google Play.

Find what your best users did differently

This is the single most useful analysis you can run. Take users who were still active after 30 days and users who left in week one, and compare what each group did in its first three sessions. There is nearly always one behavior that separates them: adding a second item, inviting someone, turning on reminders, finishing a second lesson on day two.

Once you find it, rebuild the first session to get more people to do that one thing. Then compare the next weekly cohort against the last. If retention does not move, the behavior was a symptom of motivated users and not a cause, and you test the next candidate.

Win back the ones who left

• Apple's in-app events put a card for a challenge, a new season or a major update on the App Store. People who have downloaded the app before see the event card in search results in place of your screenshots, which makes events a re-engagement tool that costs nothing.
• For subscription apps, Apple supports win-back offers aimed at people who cancelled. They work best on monthly plans: 20% of churned monthly subscribers reactivate, against 5% of annual ones.
• Google Play lets you build a custom store listing for inactive users, so a lapsed user sees what is new and not the first-time pitch.
• Google Ads has app engagement ads that target people who already have the app and deep link to a specific screen.

A weekly routine

1. Check activation rate and day 1, 7 and 30 retention for the newest cohorts.
2. Find the onboarding step with the biggest drop.
3. Change one thing about that step.
4. Check notification opt-outs and read the week's reviews.
5. Log what you changed and when.

Engagement pays twice. It raises what each user is worth, which is the subject of the guide to app monetization, and that in turn raises what you can afford to spend acquiring them, covered in the guide to app marketing. If you want software to run messaging and campaigns, the comparison of app engagement platforms lists the main options.

Mistakes that cost the most

• Showing the notification permission dialog on first launch.
• Forcing account creation before the user has seen anything.
• Building new features to fix a retention problem that starts in onboarding.
• Sending the same message to every user at the same time.
• Measuring opens, then wondering why revenue did not follow.

FAQs

What is app engagement?

App engagement describes how actively people use an app: how often they open it and whether they complete the actions it exists for. It is measured with activation rate, retention, stickiness (daily active users divided by monthly active users) and core actions per user.

How can I increase app engagement?

Start with the first session. Cut the steps between install and the core action, move sign-up after the first success, and ask for notification permission only once the user has seen value. Then compare users who stayed with users who left to find the early behavior that separates them.

What is a good engagement rate for an app?

There is no universal figure, because normal use differs by category. A messaging app is opened many times a day and a travel app a few times a year. Compare each new cohort with your own earlier ones, and with published benchmarks for your own category only.

What is the difference between engagement and retention?

Retention measures whether users come back over time. Engagement measures what they do and how often. A user can be retained with low engagement, or highly engaged for two weeks and then gone.

How do you measure app engagement?

Define the core action that represents real use, then track how many new users complete it in their first session, how many return on day 1, 7 and 30, and how many core actions an active user completes each week. Group users by install week so you can see whether changes worked.

Sources

• RevenueCat: State of Subscription Apps 2026 (day-zero conversions and cancellations, cancellation reasons and reactivation rates; medians for apps using RevenueCat)
• Android Developers: notification runtime permission
• Apple Developer: in-app events and subscriptions (win-back offers)
• Google Play Console Help: custom store listings and Android vitals